The Impact of Social Vulnerability on Exercise Outcomes: A Longitudinal Study of Physical Function in Older People With HIV

Abstract

Objectives

To examine whether the Social Vulnerability Index (SVI) moderated changes in physical function during supervised and unsupervised phases of an exercise intervention in older adults with HIV.

Methods

This exploratory analysis used data from the High Intensity Exercise Study to Attenuate Limitations and Train Habits in Older Adults with HIV. Participants completed 16 weeks of supervised high- or moderate-intensity exercise, followed by 12 weeks of unsupervised support (personalized coaching or educational text messages). Linear mixed-effects models assessed SVI moderation effects on 400-meter walk time (400-MWT) and Patient-Reported Outcomes Measurement Information System (PROMIS) Physical Function (PF).

Results

Among 117 participants, SVI did not moderate 400-MWT in either phase or PROMIS-PF during supervised exercise; however, higher SVI was associated with declines in self-reported physical function during the unsupervised phase (P = .03), robust to adjustment and sensitivity analyses.

Conclusions

Findings suggest supervised exercise benefits participants across vulnerability levels, but sustained support may be needed to prevent disparities.

Parent Trial Information

The High-Intensity Exercise Study to Attenuate Limitations and Train Habits in Older Adults with HIV trial, ClinicalTrials.gov: NCT04550676.

Plain Language Summary

How Social and Economic Challenges Shape Exercise Results

We studied whether living in a more socially and economically disadvantaged community affected how older adults with HIV responded to an exercise program. Participants took part in 16 weeks of supervised exercise, followed by 12 weeks where they exercised on their own with either coaching support or educational text messages. We looked at two outcomes: how fast participants could walk 400 meters and how they rated their own physical function. Community disadvantage did not affect improvements in walking speed during either phase of the program. It also did not affect self-reported physical function while exercise was supervised.

However, once the supervised program ended, people living in more disadvantaged communities were more likely to report declines in their physical function.

These findings suggest that supervised exercise programs benefit older adults with HIV regardless of where they live, but ongoing support may be especially important for those facing greater social and economic challenges.

Keywords

HIV social determinants of health exercise physical function social vulnerability index

Introduction

Social determinants of health (SDOH) significantly affect the health outcomes of people with HIV (PWH). SDOH are estimated to account for a substantial proportion of health outcomes at the population level, often cited as approximately 50%, whereas clinical care contributes a smaller proportion of health outcomes (20%).^1,2 These determinants, such as income, education, race/ethnicity, housing, and social support, can influence access to care, adherence to treatment, and overall quality of life.³ The impact of SDOH on health outcomes is likely amplified among older PWH because access to care and routine follow-up are crucial for consistent adherence to antiretroviral therapy, attainment of durable viral suppression, and management of comorbidities that are prevalent among PWH.^2,4–8

SDOH are key components of social vulnerability, defined as the degree to which a population is susceptible to harm from external stresses.^9–11 As individual SDOH can be difficult to measure in the context of clinical and public health research, the Social Vulnerability Index (SVI) is a tool to estimate social vulnerability on a community level.¹² The SVI is composed of factors that may disproportionately affect PWH, including socioeconomic stressors, household characteristics, racial/ethnic status, and housing type and transportation.¹³ A higher score represents greater vulnerability and less resilience to community stressors and has been strongly associated with poor health outcomes.¹³

While social vulnerability is known to affect physical function in both people with and without HIV, its role in shaping the effectiveness and sustainability of interventions targeting physical function has not been well characterized in PWH.^14,15 Evidence from other populations (eg, older adults, patients in cardiac rehabilitation, cancer survivors) shows that higher social vulnerability can limit engagement in and benefits from health interventions.^16,17 For socially vulnerable PWH, factors such as access to exercise resources, social support, and the broader environment may mediate the effectiveness of physical function interventions, namely exercise. For example, individuals living in resource-poor environments or those who face stigma or discrimination may encounter additional barriers to participating in or adhering to exercise programs.¹⁸ Understanding how these social vulnerabilities influence the outcomes of exercise interventions is essential for tailoring programs that are both effective and sustainable for PWH.

In this study, our goal was to examine how social vulnerability (measured by SVI) influences changes in physical function across both supervised (0-16 weeks) and unsupervised (16-28 weeks) phases of an exercise intervention. Additionally, we aimed to analyze whether attrition and adherence to exercise interventions differed by SVI. We hypothesized that higher social vulnerability (higher SVI scores) would be associated with smaller improvements in physical function during the supervised exercise phase (0-16 weeks) and less sustained improvement during the unsupervised exercise phase, regardless of the supervised exercise intervention arm. Further, we hypothesized that higher SVI would be associated with greater study withdrawal and lower attendance.

Methods

Study Design

This study is an exploratory secondary analysis of data from the High-Intensity Exercise Study to Attenuate Limitations and Train Habits in Older Adults with HIV (HEALTH) trial. HEALTH was a randomized, parallel-group, superiority exercise trial to evaluate the effects of supervised exercise interventions on physical function among older PWH, conducted at the University of Colorado Anschutz in Aurora, Colorado, and the University of Washington in Seattle, Washington (ClinicalTrials.gov: NCT04550676). The study protocol and primary outcomes have been published previously.^19,20 The Colorado Multiple Institutional Review Board (#19-1985) approved this study, with the University of Washington relying on this approval through a reliance agreement. Participants provided written informed consent before any research procedures. The enrollment occurred from March 2021 to August 2024. The reporting of this study conforms to the Proper Reporting of Evidence in Sport and Exercise Nutrition Trials (PRESENT) 2020 checklist (Supplemental Material S1).²¹

Participants

All participants were aged 50 years and older, reported being sedentary at baseline (not exercising 2 days/week or more), reported fatigue (≥ 2.0 on either of the first 2 screening items on the HIV-Related Fatigue Scale), had an HIV-1 RNA level < 200 copies/mL on antiretroviral therapy for a minimum of 12 months prior to enrollment, were in possession of a cell phone with texting capabilities, were able to read, speak, and write in English, and had a primary address in Colorado or Washington. Individuals were excluded if they had major contraindications to high-intensity exercise, such as severe mobility limitations, unstable angina, supplemental oxygen use, or uncontrolled hypertension.¹⁹

Intervention

After screening, participants were randomized 1:1 to high intensity interval training or continuous moderate intensity exercise for a 16-week supervised intervention. At week 16, participants began the unsupervised exercise phase (week 16 to week 28), in which they were rerandomized to a biobehavioral text-messaging support intervention delivering either personalized coaching messages or general educational messages encouraging physical activity of their choosing.²² Due to the nature of the interventions, neither participants nor study personnel responsible for delivering the supervised exercise sessions or collecting outcome data were blinded to group assignment.

Data Collection and Measures

Data were collected through a standardized questionnaire using secure web-based software (REDCap), which included demographics (age, sex, race/ethnicity, education, employment), clinical HIV characteristics (years since HIV diagnosis, years of antiretroviral therapy, current CD4 T-cell count), social characteristics (tobacco and other substance use), and health characteristics (comorbidities [yes/no categories for hypertension, diabetes, hyperlipidemia, cardiovascular disease, depression], Veterans Aging Cohort Study scores, and Patient Health Questionnaire-9 scores).^23,24

The primary outcomes were objective physical function measured by 400-meter walk time (MWT; collected at baseline, then every 4 weeks through week 28) and subjective physical function measured by Patient-Reported Outcomes Measurement Information System physical function survey (PROMIS PF, 8b, version 2.0; collected at baseline, week 16 and week 28).²⁵ 400-MWT was completed as 8 laps on a 25-m course (Colorado) or 10 laps on a 20-m course (Washington) as quickly as possible at a safe, steady pace, without running. Participants were allowed to rest while standing for up to 60 s at a time, if necessary. Time to completion in seconds was recorded.²⁶ PROMIS PF assesses self-perceived upper and lower body functionality and ability to complete activities of daily living and includes 8 items with response options on a 5-point Likert scale, ranging from 1 = “never” to 5 = “always” (a higher score indicates greater physical function) scored as a t-score, using its validated algorithm.²⁵ Body mass index (BMI) was also explored as an outcome over the supervised portion of the intervention.

Social vulnerability was assessed with the SVI and uses 16 United States Census variables from the 5-year American Community Survey, grouped into 4 themes (socioeconomic status, household characteristics, racial and ethnic minority status, and housing type and transportation).¹³ Census tract was determined by the residential address of the participant. The range of SVI is 0.0 to 1.0 (or percentiles from 0 to 100) with higher SVI scores indicating greater vulnerability.¹³

Statistical Analyses

Baseline characteristics were summarized using mean (standard deviation), median (interquartile range), or count (%) for normally distributed, nonnormal, and categorical data, respectively. To descriptively explore baseline and study participation characteristics, SVI tertiles were created by equally dividing randomized participants with SVI data into low (SVI of 1.0-49.6), middle (49.7-80.8), and high (81.0-99.8) groups.

Time was defined in terms of study weeks where assessments took place and intervention phases (supervised: weeks 0-16; unsupervised: weeks 16-28). For statistical analyses, time was modeled as a categorical variable within each phase to allow for nonlinear trends over time. Assessments of 400-MWT were conducted at baseline and every 4 weeks through week 28, while PROMIS PF was assessed at baseline, week 16, and week 28. To reflect differences in intervention conditions, separate models were fit for the supervised and unsupervised phases.

The relationship between SVI and study participation was evaluated to help characterize potential missing data patterns that could affect interpretation of primary analysis results. Exercise training attendance was calculated as the proportion of 48 expected supervised exercise sessions attended during part 1 of the study; the relationship with SVI was explored for preliminary testing through Spearman correlation. A logistic regression model was used to test whether baseline SVI predicted withdrawal during the supervised phase of the intervention, with and without adjustment for age, birth sex, site and randomized exercise intensity; withdrawal in phase 2 (unsupervised exercise) was not evaluated because only 2 participants discontinued the study in this phase.

To address our primary objective (how social vulnerability influences changes in physical function across both supervised and unsupervised phases of exercise interventions), moderation analyses treating SVI as a continuous variable were used; data from all randomized participants was included, regardless of completion of the study part (intention-to-treat approach). Linear mixed effects models were used to assess for the presence of a moderation effect of SVI on outcome trajectories over each study phase (ie, separate models for supervised exercise and independent exercise phases), pooling over randomized group in the primary approach (time × SVI interaction) and, secondarily, evaluating a 3-way interaction (time × SVI × randomized group [ie, exercise intensity in part 1 models or text support in part 2 models]). Random intercept terms (participant) were included in all models, with random slope terms (participant × time) included only if model fit improved according to the Akaike Information Criterion; time was treated as a categorical variable in models with more than 1 follow-up data collection. As interactions involving a continuous variable (SVI) can be difficult to interpret, marginal mean estimates at select SVI values, falling within the observed range in the cohort, are provided to illustrate model findings both within the text and figures. These values are model estimates for a participant living in a neighborhood with the specified SVI score (value of: 25, 50, 75 or 90% on the SVI scale).

Sensitivity analyses (1) adjusting for variables used to balance the randomization (age, sex at birth, and site), and (2) adjusted and restricted to participants completing the study phase were used to assess the robustness of findings. Sensitivity models of change over study phase 2 (unsupervised exercise) also included adjustment for the observed value of the outcome at study entry (screening). To account for potential ceiling effects, additional sensitivity analyses using Tobit modeling were performed for PROMIS physical function scores, as approximately one-third of participants scored maximum points at the baseline assessment. BMI model results and all sensitivity analyses are included in Supplemental Table S1.

All statistical testing was performed using a 2-sided test and assuming an alpha of 0.05; no adjustment for multiple comparisons was made. All analyses were conducted in RStudio (version 4.5.1, R Core Team, 2025) or Stata (version 18.5, StataCorp, 2025).

Results

Participant Characteristics

Of the 118 enrolled participants, 117 participants had available SVI data: SVI scores ranged from 1.0% to 99.8%, with more than two-thirds of participants having a SVI score of 50% or greater (N = 77). Median age was 57 years (IQR = 54-61), and median time since HIV diagnosis was 24 years (IQR = 18-30). The Moderate and High SVI tertiles were associated with greater prevalence of obesity, cardiometabolic comorbidities, and prefrailty/frailty (Table 1).

Table 1.

Baseline Characteristics of Participants by Social Vulnerability index (SVI).

		Overall (n = 117)	Low SVI Tertile (n = 39)	Moderate SVI Tertile (n = 39)	High SVI Tertile (n = 39)
SVI range		1.0-99.8%	1.0-49.6%	49.7-80.8%	81.0-99.8%
Age—median (IQR)		57.0 [54.0, 61.0]	58.0 [53.5, 62.0]	56.0 [53.5, 61.0]	57.0 [55.0, 61.0]
Sex—n (% female)		16 (13.7)	6 (15.4)	6 (15.4)	4 (10.3)
Race/ethnicity—n (%)
	African American	20 (17.1)	5 (12.8)	6 (15.4)	9 (23.1)
	Hispanic (regardless of race)	16 (13.7)	2 (5.1)	7 (17.9)	7 (17.9)
	White/Anglo, non-Hispanic	74 (63.2)	31 (79.5)	21 (53.8)	22 (56.4)
	Other	7 (6.0)	1 (2.6)	5 (12.8)	1 (2.6)
Education—n (%)
	<11 years, high school or general educational development	21 (18.0)	5 (12.8)	8 (20.6)	8 (20.6)
	Some college or technical school	44 (37.6)	12 (30.8)	19 (48.7)	13 (33.3)
	Completed college, bachelor's degree, or postcollege	52 (44.5)	22 (56.4)	12 (30.7)	18 (46.2)
Employment—n (%)
	Disabled (permanent or temporary)	27 (23.1)	8 (20.5)	8 (20.5)	11 (28.2)
	Retired	23 (19.7)	9 (23.1)	7 (17.9)	7 (17.9)
	Work part-time or full-time	50 (42.7)	21 (53.8)	15 (38.5)	14 (35.9)
	Other/missing	17 (14.5)	1 (2.6)	9 (23.0)	7 (17.9)
Current smoker—n (% yes)		13 (11.1)	1 (2.6)	5 (12.8)	7 (17.9)
Illicit drug use in the last 2 years—n (% yes)		16 (13.7)	8 (20.5)	4 (10.3)	4 (10.3)
Years since HIV diagnosis—median [IQR]		24.0 [18.0, 30.0]	23.0 [15.0, 29.5]	23.0 [18.0, 29.5]	25.0 [20.0, 30.5]
Years of antiretroviral therapy—median [IQR]		20.0 [13.0, 26.0]	17.0 [9.0, 25.0]	19.0 [12.5, 25.5]	23.0 [16.5, 26.5]
Current CD4 T-cell count (cells/mm³)—mean (IQR)		674.0 [499.8, 840.3]	593.0 [430.0, 804.0]	750.0 [582.0, 953.0]	667.5 [483.8, 784.5]
Body mass index (kg/m²)—mean (SD)		29.9 (5.9)	28.2 (5.2)	31.1 (6.4)	30.5 (5.8)
Obese body mass index ≥30—n (%)		50 (42.7)	10 (25.6)	20 (51.3)	20 (51.3)
Hypertension—n (% yes)		58 (49.6)	15 (38.5)	21 (53.8)	22 (56.4)
Hyperlipidemia—n (% yes)		63 (53.8)	15 (38.5)	22 (56.4)	26 (66.7)
Depression, anxiety, or bipolar disorder—n (% yes)		63 (53.8)	16 (41.0)	24 (61.5)	23 (59.0)
Diabetes—n (% yes)		18 (15.4)	2 (5.1)	7 (17.9)	9 (23.1)
Cardiovascular disease—n (% yes)		9 (7.7)	2 (5.1)	3 (7.7)	4 (10.3)
VACS score—mean (SD)		32.8 (9.3)	34.8 (10.8)	32.1 (9.0)	31.3 (7.6)
Frailty^b classification—n (% prefrail/frail)		63 (53.9)	14 (35.9)	25 (64.1)	24 (61.6)
Withdrew prior to second randomization		26 (22.2)	8 (20.6)	8 (20.5)	10 (25.6)
Withdrew before completing full 28-week study		28 (23.9)	9 (23.1)	8 (20.5)	11 (28.2)
Attendance during supervised exercise (%)		95.8 [85.4, 100.0]	95.8 [89.6, 100.0]	95.8 [84.4, 100.0]	95.8 [67.7, 100.0]
Baseline physical function measures
400-meter walk time (s)—mean (SD)		264.2 (50.7)	252.0 (45.6)	274.9 (60.5)	265.8 (42.8)
PROMIS physical function t-score—mean (SD)		49.3 (8.0)	51.1 (7.3)^a	47.8 (7.8)	49.0 (8.7)

Note. IQR, interquartile range; GED, general educational development; SD, standard deviation; VACS, Veterans Aging Cohort Study (VACS) 2.0 Index.

Missing data = 1.

defined by Fried et al criteria (2001).

Retention and Attendance

Study withdrawal from week 1 to 16 (supervised phase) and from week 16 to 28 (unsupervised phase) was slightly greater in the High SVI tertile (Table 1); however, there was no statistical evidence that higher SVI predicted withdrawal among randomized participants during the supervised exercise phase in adjusted or unadjusted models (adjusted OR for 1 decile increase in SVI: 1.1 [95% CI 0.9, 1.3], P = .42; unadjusted: 1.1 [0.9, 1.3], P = .40). Among participants who did not withdraw, median exercise session attendance exceeded 97% across SVI tertiles, with a weak, nonsignificant association between SVI and % attendance (Spearman correlation: −0.08).

400-MWT

In mixed-effects models of 400-MWT, there was no evidence that SVI moderated participant responses to the supervised exercise intervention (SVI × time interaction, P = .37; Table 2), with estimates of intervention response at various SVI values exhibiting similar improvement from baseline to week 16 (Figure 1; improvement in seconds between baseline and week 16 by SVI value: 25% 13.0 [95% CI 7.0, 18.9]; 75% 15.0 [10.7, 19.3]). This finding remained consistent after adjustment for age, sex at birth, and site (P = .35) and in completer-only analyses (P = .24). Similarly, no evidence of SVI moderation was observed during the unsupervised exercise phase (week 16 to week 28, SVI × time interaction: P = .54) (Table 2; Figure 1). Three-way interactions including randomized group (supervised exercise or text support) also did not reach statistical significance (Supplemental Table S2, Supplemental Figure S2).

Figure 1.

Model estimates of 400-meter walk times (MWT); (A) and change (B and C) over the intervention assuming select social vulnerability index (SVI) values (pooled over exercise or text support randomization arms).

Table 2.

Moderation of Change in 400-Meter Walk Time (MWT) by Social Vulnerability Index (SVI), Study Phase, and Model.

Study Part	Model	Effect	Model Estimate [95% CI] or Test Statistic	P-value
Supervised exercise phase (baseline—week 16)	Pooled	Effect of 1 decile increase in SVI (10%-points) on baseline 400-MWT (s)	2.1 [−1.2, 5.4]	.21
		Estimated change in 400-MWT for SVI of 50% (s; week 16 baseline)	−14.0 [−17.9, −10.0]	<.01*
		Effect of increase of 1 decile in SVI on week 16 change from baseline in 400-MWT (s)	−0.4 [−1.8, 1.0]	.55
		Categorical time x SVI interaction	F = 1.08	.37
	Exercise arm comparison	Categorical time × SVI × arm interaction	F = 1.94	.11
Unsupervised exercise phase (week 16—week 28)	Pooled	Effect of 1 decile increase in SVI (10%-points) on week 16 400-MWT (s)	1.2 [−2.1, 4.5]	.47
		Estimated change in 400-MWT for SVI of 50% (s; week 28–week 16)	3.4 [−0.7, 7.6]	.10
		Effect of increase of 1 decile in SVI on change in 400-MWT between week 28 and week 16 (s)	0.3 [−1.1, 1.7]	.68
		Time x SVI interaction	F = 0.73	.54
	Text support comparison	Time x SVI x arm interaction	F = 1.20	.31

Note. Models are intention-to-treat mixed-effects linear regressions; see supplement for results from sensitivity analyses.

PROMIS Physical Function

During the supervised exercise phase, pooled analyses showed no evidence of moderation by SVI (SVI × time interaction, P = .32; Table 3), although estimates of the change in PROMIS physical function score at an SVI of 50% and greater SVI did reach statistical significance (improvement in score between baseline and week 16 by SVI value: 25% 1.8 [−0.4, 4.0], 75% 3.0 [1.4, 4.6]; Figure 2). In models allowing for moderation by exercise intensity, SVI was differentially, albeit nonsignificantly, associated with functional change across randomized arms (3-way interaction, P = .06; Supplemental Figure S3, Supplemental Table S2).

Figure 2.

Model estimated change over the supervised exercise phase (A) and independent exercise phase (B) with 95% confidence band for select SVI values (eg, SVI of 25, 50, 75, or 90%) in PROMIS physical function t-scores. The black vertical line represents the point of no change; increases in PROMIS physical function t-scores represent improvement in self-perceived function. SVI, social vulnerability index; PROMIS, Patient-Reported Outcomes Measurement Information System.

Table 3.

Moderation of Change in PROMIS Physical Function t-Scores by Social Vulnerability Index (SVI), Study Phase, and Model.

Study Part	Model	Effect	Model Estimate [95%CI] Or Test Statistic	P-value
Supervised exercise phase (Baseline—week 16)	Pooled	Effect of 1 decile increase in SVI (10%-points) on baseline score	−0.2 [−0.7, 0.3]	.41
		Estimated change in score for SVI of 50% (week 16 baseline)	2.4 [0.9, 3.9]	<.01*
		Effect of increase of 1 decile in SVI on week 16 change from baseline in score (time x SVI interaction)	0.3 [−0.3, 0.8]	.32
	Exercise arm comparison	Time × SVI × arm interaction	3.56	.06
Unsupervised exercise phase (week 16–week 28)	Pooled	Effect of 1 decile increase in SVI (10%-points) on week 16 score	0.2 [−0.7, 0.4]	.54
		Estimated change in score for SVI of 50% (week 28–week 16)	0.0 [−1.4, 1.4]	.99
		Effect of increase of 1 decile in SVI on change in score between week 28 and week 16 (time x SVI interaction)	−0.5 [−1.0, −0.0]	.03*
	Text support comparison	Time x SVI x arm interaction	0.00	.98

Note. PROMIS, Patient-Reported Outcomes Measurement Information System. Models are unadjusted, intention-to-treat mixed-effects linear regressions; see supplement for results from sensitivity analyses.

In the unsupervised exercise phase, SVI significantly moderated changes in PROMIS PF (SVI × time interaction, P = .03); participants with higher SVI tended to experience declines in PROMIS PF, whereas those with lower SVI did not have statistically significant changes from week 16 (change from week 16: 25% 1.3 units [−0.8, 3.4]; 90% −2.1 units [−4.1, −0.1]). This moderation effect remained statistically significant after adjustment for baseline PROMIS PF score and covariates (adjusted interaction P = .03), in completer-only analyses (P = .04), and in Tobit sensitivity models accounting for ceiling effects (P = .03) (Table 2). The 3-way interaction including randomized text messaging support (personalized vs standardized) was not significant (P = .98).

Discussion

Our study provides a novel approach to understanding social vulnerability as a potential moderator of objective and subjective measures of physical function within an exercise intervention among older PWH. Overall, we found that (1) higher SVI was associated with declines in PROMIS-measured physical function (both arms) during the unsupervised phase and (2) despite slightly higher withdrawal rates in the highest SVI tertile, overall study retention and adherence did not differ meaningfully by SVI. Notably, we found that SVI did not moderate the changes in objective physical function (400-MWT) during either the supervised or unsupervised phases of the trial, or changes in PROMIS physical function scores during the supervised phase.

In contrast to what we expected, the self-reported and objectively measured physical function outcomes of participants retained in supervised exercise training were not moderated by SVI. This suggests that under ideal circumstances with professional exercise supervision and access to exercise facilities, people who are more socially vulnerable can achieve similar exercise-related health outcomes as their less vulnerable counterparts. Alternatively, because SVI is a community driven measure (ie, census), rather than an individual measure, it is possible that the social vulnerability of the enrolled participants was less than their SVI might suggest.

During unsupervised exercise we found that self-reported but not objective physical function declined among people with higher SVI. During this phase, participants were not engaged in structured group exercise and instead continued their activity independently with remote support, which may have reduced motivation and social reinforcement, both of which are known to influence adherence and performance in exercise interventions.^27,28 Prior work has demonstrated that participants in supervised exercise programs achieve better adherence and outcomes than in unsupervised settings, likely due to encouragement and group dynamics.²⁹

The PROMIS PF assesses a person's self-reported capability to complete daily activities, and our study results suggest that measured physical function may not translate to perceived abilities. Alternatively, participants with higher SVI may have had lower self-efficacy for unsupervised exercise, which led them to perceive their function was not improving. In fact, studies have implicated life stressors as the biggest contributor to perceived functional capacity in PWH, further compounded by the anxiety and depression that may accompany such events.³⁰ Exercise interventions may have a more sustainable impact among people experiencing more social vulnerability if stress management, self-efficacy development, and broader support are incorporated into the intervention.³¹

We hypothesized greater study withdrawal and lower attendance with increasing SVI, but this was not observed in our study. Admittedly, our results are limited to randomized participants, in a cohort with limited variation, and may not be upheld if data were available for individuals who declined participation or were excluded during screening. Additionally, while the observed range of SVI in this cohort spanned 1 to 100, it may still not be representative of the eligible population. Nonetheless, our finding may prompt investigators to consider the barriers that people experiencing social vulnerability may encounter when entering an exercise intervention and, importantly, how these barriers contribute to sustained lifestyle changes outside of the research setting. Costs associated with exercise, health status, and environmental factors such as neighborhood and stigma, influence exercise engagement among PWH in real-world settings.^15,32 Community-driven protocol development has been identified as a means to mitigate some barriers to retention of vulnerable populations in research studies and should be incorporated in the development of exercise interventions.³³

Strategies for long-term engagement in regular exercise can mitigate the impact of HIV on physical function in older PWH. It is well established that poor physical function in older PWH contributes to the development of comorbid conditions including frailty and disability, falls, hospitalizations, and premature death.^4,5,34–38 SDOH exacerbate predictors of poor physical function, and social support and income are the most frequently cited determinants associated with poor physical function in PWH.^27,39 Studies, including HEALTH, are experimenting with the structure of exercise interventions (mode of exercise, supervision level, duration of training) to improve health outcomes among PWH.^19,40 Additional attention must be given to the amount and type of support offered during interventions so that social vulnerability does not remain a barrier to improved health outcomes.

This study has many strengths, including the ability to report SVI within a multisite, randomized trial to improve physical function among PWH. The outcomes of this study include validated measures of physical function: the objective 400-MWT and the subjective PROMIS physical function score.^25,26 Further, to our knowledge, the impact of SVI on exercise interventions in PWH has not been reported elsewhere. There are limitations in what we report, while the study had a very large sample size for an exercise trial, we were underpowered to assess this type of relationship, and modeling approaches, chosen to maximize power in this setting, may not have allowed for appropriate complexity to approximate the true relationship. Additionally, changes in the PROMIS PF scores were limited by a ceiling effect, given that nearly one-third of participants had no room for improvement between baseline and week 16 of intervention. However, participants who were at the ceiling at baseline appeared distributed evenly across SVI tertiles, with little difference seen with Tobit modeling approaches. Further, data generated by the PROMIS PF instrument is susceptible to recall bias, social desirability bias, and changes in participants’ perceptions, which may not reflect true functional decline. Lastly, this study was among older PWH in 2 distinct cities within the United States and findings may not be generalizable to other populations.

Future work in this area will benefit from community-based participatory research strategies to engage communities with higher identified social vulnerability in the design of exercise interventions.¹⁵ The inclusion of implementation outcomes focused on feasibility, appropriateness and acceptability among users with higher SVI may help with real world application of exercise interventions studied in clinical trial settings. Methods such as key informant interviews may help identify strategies to retain participants and can shape components of an exercise intervention tailored to people with higher SVIs. Other strategies, such as a longer period of supervised exercise, incorporating stress management or different supports to limit vulnerability (exercise partners, transportation, gym memberships), or different types of exercise, can also be explored.

Conclusion

In conclusion, social vulnerability did not influence improvements in objective or self-reported physical function during supervised exercise, suggesting that structured programs can effectively support participants regardless of vulnerability level. However, during unsupervised exercise, higher social vulnerability was associated with declines in self-reported physical function, indicating challenges in sustaining benefits without structured support. Implementation strategies that address SDOH, such as extended supervision, transportation/resource assistance, and behavioral support, may be needed to sustain benefits for socially vulnerable adults with HIV. This not only has implications for future exercise interventions but may also help inform patient-centered exercise behavior recommendations within HIV primary care or more broadly.

Supplemental Material

sj-docx-1-jia-10.1177_23259582261448699 - Supplemental material for The Impact of Social Vulnerability on Exercise Outcomes: A Longitudinal Study of Physical Function in Older People With HIV

Supplemental material, sj-docx-1-jia-10.1177_23259582261448699 for The Impact of Social Vulnerability on Exercise Outcomes: A Longitudinal Study of Physical Function in Older People With HIV by Evelyn Iriarte, Kellie L Hawkins, Melissa P Wilson, Allison R Webel, Vitor HF Oliveira, Vincent Khuu, Grace L Kulik, Paul Cook, Debashis Ghosh, Catherine Jankowski and Kristine M Erlandson in Journal of the International Association of Providers of AIDS Care (JIAPAC)

Footnotes

Acknowledgments

The authors would like to acknowledge the participants of this study.

ORCID iDs

Evelyn Iriarte

Catherine Jankowski

Ethical Approval and Informed Consent

The study was approved by the Colorado Multiple Institutional Review Board (#19-1985), with the University of Washington relying on this approval through a reliance agreement. Participants provided written informed consent before any research procedures.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases, National Institute on Aging, National Institute of Mental Health, National Center for Advancing Translational Sciences, Division of Cancer Prevention, National Cancer Institute (Grant Nos. P30 DK048520, K24AG082527, R01AG066562, P30MH062512, R25MH108389, UM1TR004399, P30 CA015704, and T32 AI150547).

Declaration of Conflicting Interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: KME received grant funding from Gilead Sciences and has served as a consultant for ViiV, Merck, and Gilead (all of which were paid to the University of Colorado).

Data Availability Statement

The data supporting the findings of this study are available upon request.

Supplemental Material

Supplemental material for this article is available online.

References

Hood

Gennuso

Swain

Catlin

. County health rankings: relationships between determinant factors and health outcomes. Am J Prev Med. 2016;50(2):129‐135. doi:10.1016/j.amepre.2015.08.024

Menza

Hixson

Lipira

Drach

. Social determinants of health and care outcomes among people with HIV in the United States. Open Forum Infect Dis. 2021;8(7):ofab330. doi:10.1093/ofid/ofab330

Bassler

Ostrenga

Levitan

, et al. Redlining and time to viral suppression among persons with HIV. JAMA Intern Med. 2024;184(11):1329‐1337. doi:10.1001/jamainternmed.2024.5003

Beer

Mattson

Bradley

Skarbinski

. Understanding cross-sectional racial, ethnic, and gender disparities in antiretroviral use and viral suppression among HIV patients in the United States. Medicine. 2016;95(13):e3171. doi:10.1097/md.0000000000003171

Lagathu

Béréziat

Gorwood

, et al. Metabolic complications affecting adipose tissue, lipid and glucose metabolism associated with HIV antiretroviral treatment. Expert Opin Drug Saf. 2019;18(9):829‐840. doi:10.1080/14740338.2019.1644317

de Oliveira

Rubio

Anderle

, et al. Incorporating social determinants of health into the mathematical modeling of HIV/AIDS. Sci Rep. 2022;12(1):20541. doi:10.1038/s41598-022-24459-0

Iriarte

Cooley

Wisch

Erlandson

Ances

Jankowski

. Area deprivation index and frailty among older people with HIV. J Appl Gerontol. 2025;44(1):44‐51. doi:10.1177/07334648241262658

Elenwa

Gant

Johnson

. A census tract-level examination of HIV care outcomes and social vulnerability among black/African American, Hispanic/Latino, and white adults in the southern United States, 2018. J Community Health. 2023;48(4):616‐633. doi:10.1007/s10900-023-01191-y

Arena

Pronk

Woodard

. The influence of social vulnerability and culture on physical inactivity in the United States - identifying hot spots in need of attention. Am J Med. 2024;137(2):113‐121. doi:10.1016/j.amjmed.2023.10.025

10.

Gay

Robb

Benson

White

. Can the social vulnerability index be used for more than emergency preparedness? An examination using youth physical fitness data. J Phys Act Health. 2016;13(2):121‐130. doi:10.1123/jpah.2015-0042

11.

Mah

Penwarden

Pott

Theou

Andrew

. Social vulnerability indices: a scoping review. BMC Public Health. 2023;23(1):1253. doi:10.1186/s12889-023-16097-6

12.

Duncan

Bloomsburg

Ahmeti

. Utility of social vulnerability index in trauma: A systematic review. Injury. 2024;55(12):112016. doi:10.1016/j.injury.2024.112016

13.

Centers for Disease Control. Social vulnerability index. Accessed September 2, 2025. https://www.atsdr.cdc.gov/place-health/php/svi/index.html

14.

Prabhu

Oyaro

Wanje

, et al. Application of a social vulnerability index and its associations with physical frailty and disability in a cross-sectional study of older Kenyan women living with and without HIV. J Frailty Aging. 2024;13(4):552‐560. doi:10.14283/jfa.2024.71

15.

McCabe

Silverstein

, et al. Community-based exercise in the context of HIV: factors to consider when developing and implementing community-based exercise programs for people living with HIV. J Int Assoc Provid AIDS Care. 2017;16(3):267‐275. doi:10.1177/2325957416686836

16.

Fowler

Harmon

Tucker

, et al. The association between social vulnerability and geriatric assessment impairments among older adults with gastrointestinal cancers-the CARE registry. Cancer. 2024;130(18):3188‐3197. doi:10.1002/cncr.35390

17.

Jensen-Battaglia

Delmerico

Mohamed

, et al. Fatigue, physical function, and the context of social vulnerability for older adults with cancer. J Geriatr Oncol. 2025;16(6):102284. doi:10.1016/j.jgo.2025.102284

18.

Kitilya

Sanga

PrayGod

, et al. Perceptions, facilitators and barriers of physical activity among people living with HIV: a qualitative study. BMC Public Health. 2023;23(1):360. doi:10.1186/s12889-023-15052-9

19.

Oliveira

VHF

Erlandson

Cook

, et al. The high-intensity exercise study to attenuate limitations and train habits in older adults with HIV (HEALTH): a research protocol. J Assoc Nurses AIDS Care. 2022;33(2):178‐188. doi:10.1097/jnc.0000000000000276

20.

Kulik

Oliveira

VHF

Wilson

, et al. Comparing the effectiveness of high intensity interval training vs continuous moderate intensity exercise on physical function among older adults with HIV. Open Forum Infect Dis. 2026;13(1):ofag002. doi:10.1093/ofid/ofag002

21.

Betts

Gonzalez

Burke

, et al. PRESENT 2020: text expanding on the checklist for proper reporting of evidence in sport and exercise nutrition trials. Int J Sport Nutr Exerc Metab. 2020;30(1):2‐13. doi:10.1123/ijsnem.2019-0326

22.

Cook

Webel

Wilson

, et al. Efficacy of tailored messages for 28-week exercise sustainability in people with HIV. medRxiv. Mar 31 2026; doi:10.64898/2026.03.29.26349681.

23.

Harris

Taylor

Minor

, et al. The REDCap consortium: building an international community of software platform partners. J Biomed Inform. 2019;95:103208. doi:10.1016/j.jbi.2019.103208

24.

Harris

Taylor

Thielke

Payne

Gonzalez

Conde

. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42(2):377‐381. doi:10.1016/j.jbi.2008.08.010

25.

Cella

Riley

Stone

, et al. The patient-reported outcomes measurement information system (PROMIS) developed and tested its first wave of adult self-reported health outcome item banks: 2005-2008. J Clin Epidemiol. 2010;63(11):1179‐1194. doi:10.1016/j.jclinepi.2010.04.011

26.

Riebe

DEJ

Liguori

Magal

. ACSM’s guidelines for exercise testing and prescription. In: Kluwer. W, ed. 10th Ed ed. 2018.

27.

Safa

McClellan

Bonato

Rueda

O'Brien

. The role of the social determinants of health on engagement in physical activity or exercise among adults living with HIV: a scoping review. Int J Environ Res Public Health. 2022;19(20):13528. doi:10.3390/ijerph192013528

28.

Iriarte

Webel

Cook

, et al. What motivates older sedentary people living with HIV in the United States to participate in an exercise trial? A qualitative study. J Assoc Nurses AIDS Care. 2025;36(2):143‐155. doi:10.1097/jnc.0000000000000517

29.

Thet

Figueroa

Rodrigues

Singleton

Sivapalan

Mannheimer

. Immunological changes in HIV infected patients after supervised exercise. Open Forum Infect Dis. 2017;4(suppl_1):S222‐S222. doi:10.1093/ofid/ofx163.456

30.

Barroso

Leserman

Harmon

Hammill

Pence

. Fatigue in HIV-infected people: a three-year observational study. J Pain Symptom Manage. 2015;50(1):69‐79. doi:10.1016/j.jpainsymman.2015.02.006

31.

Stults-Kolehmainen

Sinha

. The effects of stress on physical activity and exercise. Sports Med. 2014;44(1):81‐121. doi:10.1007/s40279-013-0090-5

32.

Simonik

Vader

Ellis

, et al. Are you ready? Exploring readiness to engage in exercise among people living with HIV and multimorbidity in Toronto, Canada: a qualitative study. BMJ Open. 2016;6(3):e010029. doi:10.1136/bmjopen-2015-010029

33.

Tran

Cormier

J-C

Hopwood

, et al. Building trust for community-engaged research: recommendations from a qualitative study. J Particip Res Methods. 2025;6(2):89‐113. doi:10.35844/001c.131692

34.

Hawkins

Brown

Margolick

Erlandson

. Geriatric syndromes: new frontiers in HIV and sarcopenia. AIDS (London, England). 2017;31(Suppl 2):S137‐S146. doi:10.1097/qad.0000000000001444

35.

Zicari

Sessa

Cotugno

, et al. Immune activation, inflammation, and non-AIDS comorbidities in HIV-infected patients under long-term ART. Viruses. 2019;11(3):200. doi:10.3390/v11030200

36.

Kousari

Wilson

Hawkins

, et al. Weight change with antiretroviral switch from integrase inhibitor or tenofovir alafenamide-based to doravirine-based regimens in people with HIV. HIV Res Clin Pract. 2024;25(1):2339576.

37.

Erlandson

. Physical function and frailty in HIV. Top Antivir Med. 2020;28(3):469‐473.

38.

Erlandson

Schrack

Jankowski

Brown

Campbell

. Functional impairment, disability, and frailty in adults aging with HIV-infection. Curr HIV/AIDS Rep. 2014;11(3):279‐290. doi:10.1007/s11904-014-0215-y

39.

Xiang

. Social vulnerability and leisure-time physical inactivity among US adults. Am J Health Behav. 2015;39(6):751‐760. doi:10.5993/ajhb.39.6.2

40.

O'Brien

Nixon

Tynan

Glazier

. Aerobic exercise interventions for adults living with HIV/AIDS. Cochrane Database Syst Rev. 2010;2010(8):Cd001796. doi:10.1002/14651858.CD001796.pub3

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.21 MB